Proc. Natl. Acad. Sci.
USA
Vol. 89, pp. 8794-8797, September 1992
Biochemistry
Cloning and stable maintenance of 300-kilobase-pair fragments of
human DNA in Escherichia coli using an F-factor-based vector
(electoporaton/phyical mapping/human genome)
HIROAKI SHIZUYA*, BRUCE BIRREN, UNG-JIN KIM, VALERIA MANCINO, TATIANA SLEPAK,
YOSHIAKI TACHIIRI, AND MELVIN SIMONt
Division of Biology, 147-75, California Institute of Technology, Pasadena, CA 91125
Contributed by Melvin Simon, June 30, 1992
ABSTRACT A bacterial coning system for mapping and
analysis of complex genomes has been developed. The BAC
system (for bacterial artificial chromosome) is based on Esch-
erichia colt and its single-copy plasmid F factor. It is capable of
maintaining human genomic DNA fragments of >300 kilobase
pai-. Individual clones of human DNA appear to be main-
tained with a high degree of structural stability in the host, even
after 100 generations of serial growth. Because of high cloning
efficiency, easy manipulation of the cloned DNA, and stable
maintenance of inserted DNA, the BAC system may facilitate
construction of DNA libraries of complex genomes with fuller
representation and subsequent rapid analysis of complex ge-
nomic structure.
There is currently underway an intense effort to construct a
high-resolution physical map of each of the human chromo-
somes. Eventually, these maps will be composed of over-
lapping fragments of human DNA and will allow the direct
acquisition of DNA fragments that correspond to specific
genes. Completion of the physical map requires the avail-
ability of comprehensive libraries of DNA clones in appro-
priate vectors. Furthermore, the accuracy and efficiency of
physical mapping increase progressively with the size of the
clone firgments in these libraries. Thus, the construction of
libraries using yeast artificial chromosomes (YACs), which
permit cloning of fragments of ;500 kilobase pairs (kb),
represents a fundamental advance in our ability to generate
physical maps that order DNA over multi-megabase dis-
tances (1). However, some difficulties have been encoun-
tered with the manipulation of YAC libraries (2-4). Thus, for
example, in various libraries a fraction of the clones result
from co-cloning events; i.e., they include in a single clone
noncontiguous DNA fiagments. We describe here a bacterial
cloning system that we refer to as BACs, bacterial artificial
chromosomes. This system may provide a supplement and
alternative to the YAC system for some applications requir-
ing cloning of large fragments. The BAC system is based on
the well-studied Escherichia coli F factor. Replication of the
F factor in E. coli is strictly controlled (5). The F plasmid is
maintained in low copy number (one or two copies per cell),
thus reducing the potential for recombination between DNA
fragments carried by the plasmid. Furthermore, F factors
carrying inserted bacterial DNA are capable of maintaining
fragments as large as 1 megabase pair, suggesting that the F
factor is suitable for cloning of large DNA fragments (6).
Other bacterial systems for cloning large DNA have been
developed. For example, the system based on bacteriophage
P1 is in use (7). However, the P1 vector has a maximum
cloning capacity of 100 kb. A bacterial system based on F
factors has been reported (8). However, in this system,
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked "advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.
8794
human DNA inserts >120 kb have not been cloned and
characterized. The BAC system allows us to clone large
DNA from a variety of complex genomic sources into bac-
teria, where the DNA is stable, easy to manipulate, and
represents a single foreign DNA source.
MATERIALS AND METHODS
Construction of BAC Vector. A 97-base-pair (bp) synthetic
oligonucleotide containing nucleotide sequences for 17/SP6
promoters, two cloning sites (HindIII and BamHIf), and sites
for rare-cutter restriction enzymes (Not I, Eag I, Xma I, Sma
I, Bgl I, and Sfi I) were inserted into the single Sal I site of
pMBO131 (9). A 400-bp Sac I fragment of bacteriophage A
carrying the cosN site was then inserted into the Sac I site of
the plasmid. A 42-bp oligonucleotide having a bacteriophage
P1 loxP site was inserted at a Cla I site between cosN and the
cloning site. The resulting plasmid is called pBAC108L.
Preparation of High Molecular Weight DNA. We prepared
human DNA for cloning in agarose from MOLT-4 cells (a
leukemic cell line). Cells were harvested by low-speed cen-
trifugation and suspended in phosphate-buffered saline (PBS)
to a final cell density of 10V per ml. An equal volume of
liquefied 1% low-melting agarose in PBS was mixed with the
cell suspension and then the whole agarose/cell mixture was
poured into a 7 mm x 7 mm plastic mold as described by
Birren and Lai (10). After the mold was placed at 0C for 10
min to solidify, the agarose block was pushed out into 50 ml
of digestion buffer (1% sodium N-lauroylsarcosine/0.5 M
EDTA, pH 8.0, with proteinase K at 2 mg/ml). Cells were
digested at 500C for 2 days with one change of an additional
50 ml of lysis buffer, and the digest was dialyzed against four
changes of 50 ml of 50 mM EDTA, pH 8.0.
Partial Digestion of DNA. For HindIII partial digestion,
agarose plugs were first dialyzed extensively against 10 mM
Tris.HCl/1 mM EDTA, pH 8.0, with 50 mM NaCl, and then
equilibrated with HindIII digestion buffer (50 mM NaCl/10
mM Tris'HCl, pH 8.0/10 mM MgCl2) containing nuclease-
free bovine serum albumin (100 jug/ml) at 40C for 30 min.
Serially diluted HindUl (1-10 units) was added to the plugs
and incubated at 370C for 30 min with gentle shaking. Optimal
digestion conditions were determined by analyzing the size
range of digested DNA by pulsed-field gel electrophoresis.
Cloning Human DNA into pBAC. DNA ranging from 100 to
300 kb was excised from low-melting agarose pulsed-field
gels. The agarose slice was melted at 650C for 5 min and
digested with GELase (Epicentre, Madison, WI) by adding 1
unit ofenzyme, mixing gently, and incubating at 40"C for 1 hr.
One to three hundred nanograms of the human DNA was
Abbreviations: BAC, bacterial artificial chromosome; YAC, yeast
artificial chromosome.
*Permanent address: Department of Biological Sciences, University
of Southern California, Los Angeles, CA 90089.
tTo whom reprint requests should be addressed.
 Biochemistry: Shizuya et al.
ligated to HindIII-digested pBAC108L (molar ratio of 10 to 1
in pBAC excess) with 400 units of T4 DNA ligase at 16WC
overnight. To prevent religation of the vector, HindIII-cut
pBAC108L was treated with HK phosphatase (Epicentre).
Transformation of E. coli DH1OB cells (BRL) was carried out
by electroporation. Competent cells were prepared and elec-
troporation was performed according to the manufacturer's
instructions (Bio-Rad). After electroporation, cells were in-
cubated at 370C with gentle shaking for 45 min prior to
spreading on LB plates containing chloramphenicol at 12.5
pg/ml. Plates were incubated at 370C overnight. To identify
BAC transformants carrying human inserts, colonies were
lifted onto nylon filters (Hybond-N, Amersham) for hybrid-
ization with radioactively labeled total human DNA.
RESULTS
The BAC Vector. Many bacterial vectors based on high- to
medium-copy-number replicons often show structural insta-
bility of inserts, deleting or rearranging portions of the cloned
DNA, particularly DNA inserts of eukaryotic organisms that
may contain families of repeated sequences (11-13). There-
fore, it is difficult to clone and maintain large intact DNA in
bacteria. However, the F factor not only codes for genes that
are essential to regulate its own replication but also controls
its copy number. The regulatory genes include oriS, repE,
parA, and parB. The oriS and repE genes mediate the
unidirectional replication of the F factor while parA and parB
maintain copy number at a level of one or two per E. coli
genome. The BAC vector (pBAG) incorporates these essen-
tial genes as well as a chloramphenicol resistance marker and
a cloning segment (Fig. 1). The cloning segment includes (i)
the bacteriophage A cosN and P1 loxP sites, (ii) two cloning
sites (HindIII and BamHI), and (iii) several C+G-rich re-
striction enzyme sites (Not I, Eag I, Xma I, Sma I, Bgl I, and
XmaI / SmaI
NotI / BgII / SfE1
,/,
repE
FIG. 1. Construction of pBAC vector. The plasmid is based on a
mini-F plasmid, pMB0131 (9). A detailed description of the con-
struction is provided in Materials and Methods. CMR, chloramphen-
icol resistance marker.
Proc. Natl. Acad. Sci. USA 89 (1992) 8795
Sfi I) for potential excision of the inserts. The cloning site is
flanked by T7 and SP6 promoters for generating RNA probes
for chromosome walking, and for DNA sequencing of the
inserted segment at the vector-insert junction. The cosN site
provides a fixed position for specific cleavage with the
bacteriophage A terminase (14). The loxP site can be utilized
similarly. In this case P1 Cre protein catalyzes the cleavage
reaction in the presence of the loxP oligonucleotide (ref. 15
and unpublished result). These sites (cosN and loxP) allow
the convenient generation of ends that can be used for
restriction-site mapping to arrange the clones in an ordered
array. Restriction maps of the individual clones can be
determined by indirect end-labeling and subsequent partial
digestion (14, 16, 17).
Cloning Human DNA in the BAC Vector. The ligated BAC
DNA molecules are expected to be much larger than those
traditionally used in bacterial cloning. As one expects, trans-
formation becomes increasingly more difficult as the size of
DNA molecules increases (unpublished observation). We
have tested several procedures to transform large circular
F'lac DNA (180 kb) into different E. coli strains and found
that standard electroporation gives high-efficiency transfor-
mation, about 106 transformants per ,ug of DNA (data not
shown). The frequency of transformation with F'lac DNA is
about 1/40th that found with pBAC plasmid with no insert.
Among E. coli strains commonly used for electroporation, we
found a few strains that are capable of transformation by
F'lac DNA, and we chose DH1OB as host. The exact nature
of the genetic basis of the competence is not known.
Typically, when 2 1Ld of the ligation mixture containing 50
ng of input human DNA was transformed into highly con-
centrated competent cells (108), 100-1000 colonies appeared
on the plate. Colonies appearing on selective plates were
hybridized with whole human DNA to identify clones with
human inserts. Subsequent analysis indicated high variability
with respect to insert frequency from experiment to experi-
ment; 10-50% of the transformants carried human inserts,
depending upon the batch of the vector and insert DNA used.
Analysis of the Human Inserts. Although the amount of
DNA obtained from bacterial cells carrying the single-copy
plasmid is expected to be low, routine minipreps from 1.5 ml
of overnight culture yielded enough DNA for several exper-
iments. Twenty randomly picked clones having human in-
serts were analyzed by pulsed-field gel electrophoresis after
Not I digestion (Fig. 2A). All lanes contain the 6.7-kb vector
band and human insert DNA of various molecular weights.
The size distribution of the inserts in Fig. 2A ranges from 10
to 215 kb, with an average size of 100 kb. Fig. 2C shows an
additional five large clones ranging in size up to 300 kb, which
were obtained from separate transformations. The DNA in
the gel of Fig. 2A was transferred to a nylon filter and probed
with total human DNA. As expected, only the insert hybrid-
ized to the human DNA probe and no additional minor bands
appeared even after prolonged exposure (Fig. 2B). Deletions
occurring within the BACs in the E. coli host would yield
additional minor bands on restriction digestion. These were
not detected by ethidium bromide staining or by the more
sensitive hybridization assay. This suggests that human in-
serts on the BAC vector are structurally quite stable within
the resolution of the pulsed-field gel electrophoresis analysis.
To further examine the structural stability of such large
regions of human DNA cloned in E. coli, the restriction
patterns of DNA taken before and after long-term culture
were compared. Overnight cultures of two clones (125-kb and
175-kb inserts) were diluted and used to inoculate fresh broth
for overnight growth. This process was repeated for 5 days
(about 100 generations). The sizes of the restriction fragments
produced by digestion of the DNA from the final cultures
(Nhe I, Sac I, and Xho I) were compared with those from the
original day 1 DNA. No visible difference between the two
87% Biochemistry: Shizuya et al. Proc. NatL Acad. Sci. USA 89 (1992)

A B C

--
3315
291.0
242.5
194.0
-Wab _
145.5

FIG. 2. Analysis of human BAC clones by pulsed-field gel electrophoresis. pBAC with inserts was prepared from small cultures (1.5 ml) of
transformed colonies by a standard alkaline lysis procedure. (A) Ethidium bromide-stained BAC DNA analyzed by pulsed-field gel
electrophoresis. DNA minipreparations were digested with Not I. Separation of the digested DNA was carried out on a Bio-Rad CHEF Mapper'
apparatus for 20 hr at a field strength of 6 V/cm in a 1% agarose gel in 45 mM Tris/45 mM boric acid/1 mM EDTA at 16'C with a linear pulse
time ramping from 5 to 15 sec. (B) Southern blot showing hybridization to radioactive total human DNA probe. DNA from A was transferred
to a nylon filter (Hybond-N, Amersham). DNA hybridization was performed at 42TC in 50%o formamide for 16 hr and the filter was washed three
times with 0.3 M NaCl/0.03 M sodium citrate, pH 7.0, at 470C. (C) Sizing of five large BAC DNA. For separation of DNA fragments of 50-500
kb, switch times are ramped from 7 sec to 45 sec over 29 hr.
samples after digestion and pulsed-field gel electrophoresis the case of YACs, the size of the inserts can be larger than
was observed (Fig. 3). Moreover, Not I-digested DNA from 300 kb. However, a high percentage of YAC clones, partic-
12 individual colonies derived from the day 5 culture showed ularly higher molecular weight inserts, represent chimeric
no visible differences when compared with each other or with molecules. These chimeric clones may hamper the progress
DNA from the same number of colonies from the day 1 of mapping and analysis of chromosomes. To evaluate the
culture (data not shown). Comparable experiments with frequency of co-cloning events among the BAC clones, we
cosmid clones (35- to 45-kb inserts) in the Super-cos or have employed fluorescence in situ hybridization with BAC
Lawrist vectors showed that up to 40% of the clones rear- DNA minipreps. We have thus far analyzed 28 randomly
ranged after a shorter period of serial growth (13). picked human BAC clones (all between 200 and 300 kb) and
We have obtained BAC clones larger than 300 kb; how- identified only one clone that potentially represents a chi-
ever, this is probably not the upper limit for BAC cloning. In meric clone (data not shown). Comparable experiments with
YAC clones carrying large human DNA inserts showed that
BJ25 BJ6 a substantial number of clones mapped to multiple sites,
I-I suggesting that many of the clones were chimeric (2). The
co-cloning frequency in the BAC system appears to be
2 ~ cu
XO
4--
)
cm
XO
-c-
significantly lower than that in the YAC. Another possible
source of chimeric YAC clones is that multiple artificial
-Z

chromosomes may be found in a single yeast cell (3, 4). In the

BAC system, the two genesparA and parB are involved in the
exclusion of extraneous F factors (two BACs in a single cell).
In one experiment we analyzed more than 300 clones and
have not found multiple BACs in a single cell.,

FIG. 3. Comparison of restriction fragments of two clones before
and after 100 generations. Two clones carrying BACs of 125 kb
(BJ25) and 175 kb (BL6) were grown in 5 ml of LB containing
chloramphenicol (12.5 ,ug/ml) and grown at 37°C overnight (day 1).
Cultures were diluted 100-fold and 0.1 ml of this dilution was
inoculated into 500 ml of LB with chloramphenicol (12.5 Pg/ml) and
incubated overnight. Serial cultures were made in this way for an
additional 3 days. Plasmid DNAs were prepared from day 1 and day
5 cultures, digested with three different restriction enzymes (Nhe I,
Sac I, and Xho I), and analyzed by pulsed-field gel electrophoresis
(day 1, left lane; day 5, right lane).
DISCUSSION
Several lines of evidence suggest the utility of the BAC vector
as a general cloning vehicle to construct comprehensive
libraries of higher organisms. (i) Large plasmids can be
transfected to E. coli by electroporation. The transformation
of bacteria by electroporation is 10-100 times more efficient
than yeast spheroplast transformation. This reduces the
amount of DNA necessary for library construction. This
feature is particularly suited to creating specialized libraries
where the source DNA is limited. (ii) Traditional bacterial
colony lifts and hybridization methods can be used for library
screening even though only a single copy of pBAC plasmid
exists in cells. (iii) Unlike YACs, which are linear, BACs with,
inserts, like large F' factors, exist as supercoiled circular
plasmids in E. coli. This permits easy isolation and manipu-
lation of the large DNA in solution with minimal breaking,,
whereas yeast chromosomes are more difficult to isolate
intact since the linear DNA is more susceptible to shear.
Furthermore, the ability to visualize the cloned DNA by
ethidium bromide staining in pulsed-field gels vastly in-
creases the speed of structural analysis of the clones. (iv)
DNA from minipreps is easily reintroduced into bacterial
cells by the same electroporation procedure. (v) A set of
 Biochemistry: Shizuya et al.
genes for F' maintenance is sufficient to sustain large foreign
DNA in the BACs. (vi) A low frequency of co-cloning in the
BACs has significant advantage in physical mapping by
"bottom-up" approaches.
One of the important aspects in any cloning system is the
genetic stability of cloned DNA. Results showing a stable
structure for human DNA in the BAC vector suggests that the
cloned DNA should represent a stable source of specific
fragments. However, more experiments including long-range
restriction-site mapping with BACs will be required to dem-
onstrate that cloned DNA fragments are indeed faithful
copies of the source DNA. Some clones in any genomic
library may be expected to show instability in a particular
host because of the recombinogenic activity of certain spe-
cific DNA sequences. A bacterial system may offer advan-
tages in this respect because the pathways of genetic recom-
bination in E. coli have been described in detail, and the use
of appropriate bacterial mutants may reduce instability prob-
lems. For this purpose we have constructed several candidate
repository strains that lack many recombination functions. A
set of new strains are based on DH1OB. They are capable of
transfection of the large DNA and deficient in host-controlled
restriction (hsdMRS), methylation-sensitive restriction
(mcrABC, mrr), and various types of general recombination
systems (recBC, sbcBC). Among the strains constructed are
HS986, having a recJ mutation, and HS979, having a recA
deletion mutation, as well as the mutations described above.
In any cloning system, analysis and fingerprinting of many
clones with large inserts require rapid and convenient meth-
ods. One fingerprinting strategy that takes advantage of the
vector involves the use of A terminase to generate cohesive
ends at the cosN site, which may then be used as reference
points to establish a restriction map by partial digestion. The
restriction pattern of each clone can be compared and used
to recognize overlapping clones. Independent labeling of
both allows us to carry out the fingerprinting analysis with
greater speed and accuracy. The other benefit of having cosN
on the vector is that the BAC vector can be used as a highly
efficient cloning vector for DNA of 40-kb average size after
packaging and infection as in conventional cosmid cloning.
Extensive libraries of fosmid (F-based cosmid) clones are
readily constructed and offer increased insert stability (13).
The clones can be easily analyzed and used in physical
mapping.
Proc. Natl. Acad. Sci. USA 89 (1992) 8797
Recently, we have begun construction of a BAC library of
total human DNA. In addition we have prepared a chromo-
some 22-specific fosmid library with 7-fold redundancy (un-
published results). The library together with fosmids will
augment physical chromosome mapping with YACs and with
conventional cosmids.
We thank Drs. M. O'Connor and W. Bender for providing the
pMBO131 plasmid prior to publication. This work was supported by
a grant from the Department of Energy, DE-FG03-89ER60891.
1. Burke, D. T., Carle, G. F. & Olson, M. V. (1987) Science 236,
806-812.
2. Green, E. D., Riethman, H. C., Dutchik, J. E. & Olson, M. V.
(1991) Genomics 11, 658-669.
3. Wada, M., Little, R. D., Abidi, F., Porta, G., Labella, T.,
Cooper, T., Della Valle, G., D'Urso, M. & Schlessinger, D.
(1990) Am. J. Hum. Genet. 46, 95-106.
4. Bellis, M., Gerard, A., Charlieu, J. P., Marcais, B., Brun,
M. E., Viegas-Pequignot, E., Carter, D. A. & Roizes, G. (1991)
DNA Cell Biol. 10, 301-310.
5. Willetts, N. & Skurray, R. (1987) in Escherichia coli and
Salmonella typhimurium: Cellular and Molecular Biology, ed.
in chief Neidhardt, F. C. (Am. Soc. Microbiol., Washington),
Vol. 2, pp. 1110-1133.
6. Low, K. B. (1987) in Escherichia coli and Salmonella typhi-
murium: Cellular and Molecular Biology, ed. in chief Neid-
hardt, F. C. (Am. Soc. Microbiol., Washington), Vol. 2, pp.
1134-1137.
7. Sternberg, N. (1990) Proc. Natl. Acad. Sci. USA 87, 103-107.
8. Hosoda, F., Nishimura, S., Uchida, H. & Ohki, M. (1990)
Nucleic Acids Res. 18, 3863-3869.
9. O'Connor, M., Pfeifer, M. & Bender, W. (1989) Science 244,
1307-1312.
10. Birren, B. & Lai, E., in Pulsed Field Gel Electrophoresis: A
Practical Guide (Academic, San Diego), in press.
11. Wyman, A. R., Wolfe, L. B. & Botstein, D. (1985) Proc. Natl.
Acad. Sci. USA 82, 2880-2884.
12. Ishiura, M., Hazumi, N., Koide, T., Uchida, T. & Okada, Y. J.
(1989) J. Bacteriol. 171, 1068-1074.
13. Kim, U.-J., Shizuya, H., de Jong, P. J., Birren, B. & Simon,
M. I. (1992) Nucleic Acids Res. 20, 1083-1085.
14. Rackwitz, H. R., Zehetner, G., Murialdo, H., Delius, H., Chai,
J. H., Poustka, A., Frischauf, A.-M. & Lehrach, H. (1985)
Gene 40, 259-266.
15. Abremski, K., Hoess, R. & Sternberg, R. (1983) Cell 32,
1301-1311.
16. Rackwitz, H. R., Zehetner, G., Frischauf, A.-M. & Lehrach,
H. (1984) Gene 30, 195-200.
17. Kohara, Y., Akiyama, K. & Isono, K. (1987) Cell 50, 495-508.